API Reference — UX Master | Complete API Documentation

CLI API

Command-line interface for UX Master.

uxm init

Install UX-Master skill for AI assistants.

Bash

uxm init --ai <platform> [options]

Options

Option	Description	Default
`--ai, -a`	Target AI platform (required)	-
`--global, -g`	Install to global config	false
`--force, -f`	Overwrite existing files	false

Supported Platforms

claude - Claude Code
cursor - Cursor IDE
windsurf - Windsurf IDE
vscode - VS Code MCP
figma - Figma Plugin
all - All platforms

Example

# Install for Claude Code
uxm init --ai claude

# Install globally with force
uxm init --ai claude --global --force

uxm search

Search UX-Master knowledge base.

Bash

uxm search <query> [options]

Options

Option	Description	Default
`--domain, -d`	Search domain	-
`--stack, -s`	Technology stack	-
`--max-results, -n`	Max results	3
`--design-system, -ds`	Generate design system	false

uxm validate

Validate UI against 37 Design Tests.

Bash

uxm validate <target> [options]

Test Suites

all - All 37 tests
mobile - Mobile-specific (7 tests)
landing - Landing page (6 tests)
dashboard - Dashboard (6 tests)
a11y - Accessibility (5 tests)

Output Formats

rich - Terminal with colors
json - JSON output
markdown - Markdown report
html - HTML dashboard

Example

# Validate HTML file
uxm validate index.html --suite mobile

# Validate URL
uxm validate https://stripe.com --url --suite all

# Generate HTML report
uxm validate index.html --format html --output report.html

uxm extract

Extract design system from website.

Bash

uxm extract <url> [options]

Example

# Extract to JSON
uxm extract https://stripe.com --output stripe.json

# Export to Figma
uxm extract https://linear.app --figma

# Deep crawl with screenshots
uxm extract https://vercel.com --depth 2 --screenshots

MCP Server API

Model Context Protocol server for AI assistants.

Base URL: http://localhost:3000

Authentication: None required for local development

Available Tools

search_ux_laws

Search 48 UX Laws for psychology-driven design.

POST /mcp/v1/tools/call

{
  "name": "search_ux_laws",
  "arguments": {
    "query": "mobile touch targets",
    "product_type": "mobile",
    "max_results": 5
  }
}

validate_design

Validate UI code against 37 Design Tests.

POST /mcp/v1/tools/call

{
  "name": "validate_design",
  "arguments": {
    "html": "<button style='width: 100px'>Click</button>",
    "css": "button { padding: 10px; }",
    "test_suite": "all"
  }
}

extract_design_system

Extract design system from website using Harvester v4.

POST /mcp/v1/tools/call

{
  "name": "extract_design_system",
  "arguments": {
    "url": "https://stripe.com",
    "depth": 1,
    "include_screenshots": false
  }
}

Python SDK

Validation Engine

Python

from validation_engine import ValidationEngine

# Create engine
engine = ValidationEngine()

# Load harvester data
with open('harvest.json') as f:
    data = json.load(f)

# Run validation
report = engine.validate(data, test_suite='all')

# Check results
print(f"Score: {report.score}/100")
print(f"Passed: {report.passed_count}/{report.total_count}")

Search Engine

Python

from search_engine import SearchEngine

engine = SearchEngine()

# Search UX Laws
results = engine.search(
    query="mobile touch targets",
    domain="ux-laws",
    max_results=5
)

# Generate design system
system = engine.generate_design_system(
    query="fintech dashboard",
    project_name="MyApp"
)

Validation Engine

Test Categories

Mobile Tests (7)

ID	Name	Severity	Check
DT-MOB-001	Fitts's Law	Critical	Touch targets ≥ 44px
DT-MOB-002	Thumb Zone	High	Actions in bottom 25%
DT-MOB-003	Touch Feedback	High	Hover/active states

Landing Page Tests (6)

ID	Name	Severity	Check
DT-LND-001	Hero Clarity	Critical	Value prop in 5 sec
DT-LND-002	CTA Prominence	Critical	Visual dominance
DT-LND-003	Social Proof	High	Above the fold

Accessibility Tests (5)

ID	Name	Severity	Check
DT-A11-001	Focus States	Critical	Visible focus
DT-A11-002	Input Labels	Critical	Associated labels
DT-A11-003	ARIA Usage	High	Proper attributes

Harvester API

Browser Injection

JavaScript

// Inject harvester_v4.js into browser console
const harvester = document.createElement('script');
harvester.src = 'https://ux-master.todyai.io/harvester_v4.js';
document.head.appendChild(harvester);

// Run extraction
const data = window.UXMasterHarvester.extract();
console.log(JSON.stringify(data, null, 2));

Python API

Python

from harvester_browser import BrowserHarvester

harvester = BrowserHarvester()

# Harvest single page
result = harvester.harvest("https://example.com")

if result["success"]:
    data = result["data"]
    print(f"Extracted {len(data['components']['blueprints'])} components")

Output Format

JSON

{
  "_version": 4,
  "meta": {
    "url": "https://example.com",
    "title": "Page Title",
    "pageType": "dashboard",
    "timestamp": "2024-01-01T00:00:00Z"
  },
  "visualAnalysis": {
    "colors": {
      "semantic": {...},
      "neutrals": {...}
    },
    "typography": {...},
    "layout": {...},
    "spacing": {...}
  },
  "components": {
    "blueprints": {...}
  }
}

Error Codes

Code	Meaning	Resolution
400	Bad Request	Check parameters
404	Not Found	Check URL/file path
422	Validation Error	Check input format
500	Server Error	Check logs
503	Service Unavailable	MCP server not running

🚀 Ready to build?

Start with our tutorials or explore the Harvester v4 documentation.

Browse Tutorials → Harvester v4 Docs